Bug in BiKalmanFittedElectrons affecting Hlt2LowMultDiElectron CEP line for 2017-18.
This is report of a bug found by @crsanche . As it affects data already taken it cannot be fixed but should still be documented.
CEP J/psi->ee events are triggered by Hlt2LowMultDiElectron https://gitlab.cern.ch/lhcb/Hlt/-/blob/2018-patches/Hlt/Hlt2Lines/python/Hlt2Lines/LowMult/ElectronLines.py They use BiKalmanFittedElectrons which rely on Brem photons https://gitlab.cern.ch/lhcb/Hlt/-/blob/2018-patches/Hlt/Hlt2SharedParticles/python/Hlt2SharedParticles/TrackFittedBasicParticles.py . If no photons are present in the event the sequence is stopped and no electrons are made, even if one could have electrons with no bremsstrahlung (Check for instance listProperties(tck,'.*/Hlt2LowMultDiElectronFilterSequence') for tck = 0x215517A7). The effect is that on simulated CEP J/psi->ee the efficiency is about 25% for no-bremsstrahlung corrected electrons.
I presume this bug affects all dielectron lines, but the likelihood of having zero photons is negligible for non-CEP data.
cc those involved at the time @johndan @graven @gligorov @rmatev @cofitzpa @apearce @sstahl For CEP: @bursche @heli @mvesteri @maxime @pgandini